LATIDIA · Ciberseguridad

QuasiLeak-Bench: Evaluación de detectores de privacidad basados en LLM en la recuperación de grupos de cuasi-evidencia condicionada por políticas

arXiv: 2610.21333v1Announce Type: cross Resumen: Los mensajes de lenguaje natural a menudo contienen cuasi-identificadores que son débiles de forma aislada, pero pueden revelar conjuntamente los atributos privados de un individuo. Llamamos a tal rev conjuntamente

La noticia

arXiv:2610.21333v1 Announce Type: cross Abstract: Natural language messages often contain quasi-identifiers that are weak in isolation but can jointly reveal an individual's private attributes. We call such a jointly revealing set of cues a quasi-evidence group. Existing privacy benchmarks focus on explicit identifiers, attribute inference, or task-driven disclosures, leaving inference from quasi-evidence groups largely unmeasured. In practice, whether exposing such a quasi-evidence group violates privacy also depends on sender-specified disclosure policies. We introduce QuasiLeak-Bench, a benchmark for policy-conditioned quasi-evidence group recovery, which evaluates whether detectors recover complete quasi-evidence groups and flag the disclosures the given policy prohibits. To obtain reliable ground truth at scale, a reverse-construction pipeline first selects weak